Anirban Basu

USING RACE IN CLINICAL ALGORITHMS Started during the heights of pandemic, this paper has been a labor of love for me for the last 3 years… finally out at @ScienceAdvances @UW_Pharmacy #UWCHOICE science.org/doi/10.1126/sc… Some discussions… 🧵 1/

@InfectDiseaseAd @UW_Pharmacy .@KyuQLee

From @Basucally and @UW_Pharmacy: #Oseltamivir is the leading outpatient #influenza #antiviral, whereas #peramivir is increasingly favored for inpatients, highlighting clinical alignment of antiviral selection with specific care settings. infectiousdiseaseadvisor.com/news/oseltamiv…

There goes the demise of log-OLS regressions..

Congratulations to Beth Devine, PhD, PharmD, MBA, MSc, as she moves into the role of ISPOR President (2026-2027). #HEOR #healthcare #leaders #leadership ow.ly/UYZC50Zhj3T

Now accepting submissions for the 11th Workshop in Behavioral and Experimental Health Economics (Dec 7-8, University of Washington, Seattle) @UWCHOICE. Deadline: Aug 31, Submit here: bit.ly/4xQZ5G6 #BehavioralEcon #HealthEcon @cawley_john @Basucally @DanielWiesen1

@4ntonioR @jondr44 @instrumenthull @NeumarkEcon @arindube @attilalindner @benzipperer These seem to be substantially different across outcome specifications! Again I have hard time interpreting what changes in E(log(emp/pop)) is showing me.

@jondr44 @Basucally @instrumenthull @NeumarkEcon @arindube @attilalindner @benzipperer We also find more negative effects with log(emp/pop) and a wider gap for unweighted (with the sign flipping). There is re-scaling when we present our estimates (as in CDLZ), but it does seem we obtain larger gaps. Here is our relevant table:

A few thoughts on this interesting new minimum wage paper by @NeumarkEcon and @4ntonioR renalyzing the QJE paper by @arindube-@attilalindner-@benzipperer, which relates to one of my favorite topics: event-studies and parallel trends! 1/

@jondr44 @instrumenthull @NeumarkEcon @4ntonioR @arindube @attilalindner @benzipperer The estimands identify different parameters I think, even though the estimates may be similar in this case. I agree Emp/Pop does not make sense , and Emp, adjusting for Pop, is better option (@jmwooldridge Possion /binomial discussions)

@Basucally @instrumenthull @NeumarkEcon @4ntonioR @arindube @attilalindner @benzipperer PT for log(Emp) and log(Pop) seemed more natural to me than PT for Emp/Pop. And once the first two were in logs it seemed natural to have log(Emp/Pop) which is mechanically the differene btwn the first two. Empirically log(Emp/Pop) vs Emp/Pop doesn't seem to matter much

@btshapir Dribing?

There should be a new vocab word for vibe coding + drinking. Kind of like drunk dialing, but for research. Thoughts?

@anup_malani @cremieuxrecueil pubmed.ncbi.nlm.nih.gov/25966453/

@cremieuxrecueil I think Tomas Philipson has a paper that suggest a good all in measure of drug benefit is relative attrition / noncompliance. Bart Hamilton and I wrote a similar empirical piece many years ago, but did not publish it.

This is a hilariously disappointing result: 15% of placebo GLP-1 patients seemingly realized they weren't getting the real stuff and so they went out and got it on their own. They're undermining the trials because they want to lose weight!

@Jabaluck What do you think will be in the utility function of a future AGI?

The idea of a future where humans retain control over increasingly powerful AIs that act as benevolent advisors seems to me unlikely to last very long. It presumes the wrong notion of intelligence, where an AI is your smart friend, as opposed to an entity that is far more capable than even a large group of coordinated humans at pursuing objectives. The very smartest humans are not a good guide to the capabilities of entities that can instantiate themselves millions of times, master every discipline down to a superhuman level of detail, build innumerable robots, and do the work of billions of people in seconds. To me, the only viable "alignment" future is one where humans are disempowered but the vastly more powerful intelligences that exist remain invested in human flourishing because: a) It has non-zero weight in their overall objective and b) It can be achieved at relatively low cost given their vast computational resources. We don't currently know how to ensure a).

New paper by @Basucally! How much of a drug's social value should go to innovators? 100% maximizes long-run welfare, but less if society prioritizes consumer surplus—especially when small countries can free-ride on others' R&D. journals.uchicago.edu/doi/10.1086/74…

Call for papers! We are Excited that Partha Deb at CUNY has agreed to host the annual health economics workshop - Oct 1-3. Please send cool papers to: AHEW.submissions@gmail.com

@ashishkjha I do agree with you that private insurance market does not work in bringing down healthcare prices..

@ashishkjha @ashishkjha .. unfortunately the price masks a lot of quantity (and some quality) aspects that’s not captured by just the procedure. Although I think we should be comparing US public reimbursements to compare with these other countries..

If you ever wonder whether we have a utilization problem or price problem in US healthcare This chart should help Here are four procedures by rates and prices, in the US, Australia, Switzerland and Germany We don't use more healthcare We just pay much higher prices

@BanilowMary @instrumenthull I am fully aware that he does. I am just surprised they even the statistician did not bring up this literature when criticizing the Poisson model..

Sigh

@jmwooldridge One area where I am unsure about this approach is when we want to understand the effect of an X on Y, and X affects both Y and POP.. then I don’t want to adjust away the channel of effects through POP….

It seems that there needs to be more consistency in models with a count y. If y clearly increases with population size -- such as a homicide count -- an exponential model for y with log(pop) as an explanatory variable makes the most sense. This allows y/pop to vary with pop.

@PhD_Genie Someone published my analyses.

Write an academic horror story in six words or less, based on real-life experiences

@KhoaVuUmn @haroldpollack Reverse causality?

Meeting my counterfactual in real life.

Residency applications to programs in abortion-restricted states declined among both women and men post-Dobbs, with the largest gaps in abortion-related specialties such as #OBGYN and family medicine. ja.ma/40KSNZj